Support Driven Expo 2023: Day 1 Notes

Notes from Support Driven Expo day 1 talks and workshops that I attended.

A Public Handbook: Building trust with the team, customers, and beyond

If you want a full write-up, go see the blog post transcript of my presentation.

The Power of Automation in Customer Support

Kim Everett, Outschool

Single system automation example: Slack workflows can help organize and operate standardized processes or reminders. Workflows can tag others, post in channels, collect information, and wait for teammates to take actions
Multi-system automation example: webhook from intercom to router to intercome/Slack/Sheets. Tracks internal notes and automate triggers other processes based on the content. One trigger initiates multiple workflows.
Transitioning to automations: repetitive tasks, what’s the ideal state, best place to save time
makes sure not affecting workflows, it’s just happening without changing how they work
Connecting the dots: map out the basic framework with what should trigger and the end results, identify trigger options, conditional logic with filters to create safeguards and allow customization
Example: Automating replies to automated emails: responding to automated reminder emails, filter to ensure user meets criteria for users to receive the initial automated reminder, result: automatically responded to over 8% of all inbox volume, over 12% of support volume
potential pitfalls: over-automation, data/privacy concerns, technical glitches/errors, infrequent maintenance
Best practices: prioritize the customer experience, start small and iterate, keep the human touch, seamless integration with tools, monitor and adjust automations, label diagrams and steps, create a feedback loop
turn on for 10% before going to 100%
document workflows with images, labels

Designing Great Conversations for Customer Success

Declan Ivory, Intercom

about a a conversation
exceptional customer support is a necessity
customers expect more than just answers, meaningful interactions and valuable relationships
must go beyond traditional customer support methods and re-think customer journey
expectations in a conversation: engagement feels natural, know who they are, context maintained, feel personalized, knowledge imparted and subject matter expertise demonstrated, “value adding” component: wow-factor
why difficult? multiple channels: app, website, chatbot, human; slack design/consideration for end to end journey
typical flow: AI bot, automation, human support
typical problems: handovers aren’t seamless, context not maintained, interactions don’t feel personalized, whole experience feels like friction
if AI bot answers the question or solves the problem then at least speed of resolution is good, but added more friction if it doesn’t
be intentional about designing the conversation flow from customer perspective
constantly measure and review
tune and adjust how the conversation flows
give it a personality (give it a name), make sure represents tone of brand/company, leverage reasoning capabilities of LLM (ask clarifying questions, make it a real dialogue), demonstrate you know the customer
use AI to summarize the conversation to date
leverage data: CRM, previous case, usage, etc.
think beyond the current issue
review CX: CSAT, CES, NPS, internal quality; focus a lot on agent performance, but need to measure customer experience across full journey of support interaction, understand every part and overall
AI Bot capabilities are evolving and changing based on underlying evolution of LLMs, with use of better and more diverse content sources
customer expectations are also evolving: what is a ‘wow’ experience today may just be ‘ok’ tomorrow
use customer measure and feedback to identify opportunities
complexity of customer journey: self-serve using knowledge/content, community > AI Bot/human, omni-channel > AI Bot/human, proactive support: automation/AI/ML/customer context data/ seamless platform

Digging In: Product discovery in the queue

Elyse Mankin, Help Scout

Support black hole: conversations with customers know what the problems are, but often feels like advocacy goes unresponded
Discovery in the queue is the process of turning raw data from your support queue into actionable insights for the business.
improves customer experience, build relations between the different teams
actionable insights: specific recommendation, support business goals, measurable impact
ambiguity or subjectively know something needs to be solved, or something is too big; need to understand what is the actual problem?
have to prioritize, make sure advocating the things that can best help the company achieve its goals
Importance: support is mission critical, improve customer’s experience with product and support, improve support’s experience with product and other teams, drive strategic results, create valuable partnerships across the org, elevate role/impact of support
start small: lean on the features in your ticketing system (tags, custom fields, reports, etc.)
non-negotiables for great impact: track key customer/business data points, understand business goals and strategies
iteration is your friend, don’t have to be perfect, don’t have to re-invent the wheel
track customer conversations: bug reports, feature requests, support conversations, surveys/NPS
dig deeper: understanding customer motivation, uncover problem they are trying to solve: expectations? value bring their team? leverage saved replies for a consistent internal process
make your case: go beyond ticket count and tell the customer’s story, combine numbers and narrative as a powerful combination, trends over time especially for prioritization, contextualize with data the business cares about
report on impact: select metrics to demonstrate impact, benchmark the data to show before and after
example: 9 hours a week save on manual work, 15% decrease in volume by improving docs, 20% trial to paid due to onboarding product improvement
scale as your grow: make data accessible to improve prioritization that engineering/product uses, business intelligence platform: cross-referencing important data points
dedicated role / growth path – clear ownership, added specialization
leverage AI/ML reporting platforms
internal alignment:
- support: have a tangible impact, why are we doing this, simplicity is best, show your progress / having the conversation to drive things for customers
- support leadership: investment in CX, clear path to tangible business value (choose certain metrics), smarter team scaling (reduce repetitive tasks/work), develop specialized growth paths (able to keep and recruit top talent)
- product leadership: help solve right problem in the right way at right time, minimize overhead (of collecting customer feedback), foundation of product discovery, signal vs. noise / which ones are important and which ones are not
tools: help desk (Help Scout), JIRA, Looker to create customer dashboards

Empowering support teams for success during product releases

Annabelle Nichols, Help Scout

even when it’s good change, change is hard
story: software company based on usage, wanted to update usage but meant billing changes, support had 2 days, working off of a screenshot and a short FAQ, breaking changes, solution was pay more, product didn’t know how customers were using product
get support involved in the changes from the beginning
pick a point person to lead the charge through the change
not one person need to do all the work, but a project manager
carve out 4-8 hours a week
size the problem: remember, you are the customer expert, trust your gut on the amount of impact
get involved early, be part of the conversation from the start
go to the team and ask: can I join your conversations?
helps to reduce the assumptions that get made
will you come talk to us too? Have a lead changemaker talk directly to your team; prep with questions: why design choice? why targeting certain customers? what’s next? provide context
Cheat sheet: secret to success
internal documentation to change product information to support, keep it 1 page for searchability, not done, keep updating with new questions and information, finish initial version 2-4 weeks ahead of launch
use the same template, same information: quick info/above the fold, important links, rollout dates, what’s changing, what’s next, why this, macros/saved replies, bugs, feature requests, troubleshooting, FAQs, training
often becomes source of truth
getting/sharing feedback: change management does not end on launch day
working with whoever is most receptive: this is the story we’re telling product team
survey post-release: what can we do to help your job
product analysts are embedded in each group
metrics: no more than 10% of support interactions after release related to bugs

10x your In-Product Support Experience

Simon Rohrbach, Plain

context: designer, a few years ago was part of deliveroo, support was a large part of the company, dealing with a lot of hangry customers, the biggest problem was not getting enough information, got interested in internal/support
chat: low friction, great for pre-sales, but difficult to get structured information, expectation of instance support, hard to manage volume, often feels separate from product, can feel like fighting chatbots to talk to a human
email: low friction, easy to use outside of product, but no structured information, no barrier: anyone can use it for any question
contact forms: structured, can be used anywhere in the product, great for routing, but old fashioned – true?
The more structured the query, the better
Can be even greater in technical B2B support
Support is not part of the product. Is it the product.
can make a contact form much better, smarter
can be floating, part of navigation, at points of friction example: when get an error
floating: partially covers UI, hard to write long messages, feels ethereal, easy to dismiss, feels separate from product
let’s go big: take up whole right column: space communicates importance
can link to getting started guides, documentation, knowledge base,
want to have: something for non-product questions (account, pricing, etc.), bug report, feature request form (may go directly to product team)
design hint: reduce choice and always have default action
example: bug report form: want to collect: who is it, what’s the issue, exact part, what happened, steps to reproduce, is it urgent, screenshot
sign in: show name, email will reply to, product area affecting, which project is affected, what’s happening, how reproduce, screenshot (if try to submit without, tell why), urgency: “This bug is preventing me from using this product.”, reply time with submit button
design hint: mirror info have back to the user
break down free text as much as possible
design hint: use product data to triage
design hint: explain what you need and why
design hint: match the user’s mental model
set expectations up front
little smarter: connect to incident/status page, check version of app, plan limitations
incorporate live data wherever we can
give best customers a little special treatment: note “Enterprise” user with lower response time
visualize the value customers are getting
examples: arc.net, linear.app, vercel.com
can usually do with a day’s worth of engineering, some product buy-in

Lunch

Watchable Wildlife-First Place-Stacey Reid, Flying Squirrel, in Waterloo, Alabama by Outdoor Alabama. CC-BY-NC-ND 2.0

Customer-Centric Compassion: Enhancing Customer Support Experiences Through Empathy

Emma Charles, Everbridge

why empathy matters: job satisfaction, effective problem resolution, personal fulfillment, reduced stress (better interaction), improved comms skills, professional growth
benefits: customer retention, positieve reputation, employee staifsaction/team morale, increased cross-selling and upselling
practice strategies: active listening (let customer know understand, read back), empathy training (not a lot, but exist in the UK), emotional intelligence, encouraging a culture of empathy, putting yourself in their shoes
questions to ask:
1. who are we empathizing with?
2. what do they need to do?
3. what do they see?
4. what do they say?
5. what do they do?
6. what do they hear?
7. What do they think and feel? pains / gains, what other thoughts and feelings might motivate their behaviour?
strategies for cultivating empathy, empathy in support leads to improved customer satisfaction, fostering a culture of empathy within the workplace not only enhances employee morale, but also sets the stage for long-term customer success.
Empathy isn’t just a soft skill, it’s a strategic asset in the world of support.
burnout on repetitive issues: reinforcing patience, might be a new customer
setting expectations: don’t apologize, stand and smile, changes tone, reading out loud written comms, working it out as a team, grammarly
survey monkey: asked customer how they’re feeling in the support form
empathy internally: gone out for drinks, mix teams up, online playing games, donut to build connections; ways to make the improvements

Strategies for Evolving Support

Valmir Verbani, Tyk

engineering background, roles in customer services and sales, current role: global head of support; centred around engineering and support
stages of friction/areas of improvement: onboarding, client frustration, support frustration
look into tickets, and analyze to see what might need change
example: pending to auto-close; 7 day pending, reopens
before closing: identified root cause, resolve issue and fully understand, customer understand how to resolve and/or next steps, customer has confirmed ticket can be closed, added RCA to internal notes, made improvement to product, doc, or cx
data: > 150 auto-closed, 63 replied that solved, 8 re-created, 64 no reply (4 attempts), 26 closed relied on internal tickets, > 10 new doc changes/additions, 5 left good ratings
spotting trends, documentation improvements, make external/internal knowledge base articles consistent
ticket ownership: enforced, advocate for client
can anything be automated?
AI: potential solution, resources to compile solution, trends
automations: example: reports
trialist experience: not deflecting free, get information about company and pass through AEs
connection between support and sales
key takeaway: always room for improvement, just revisit the data

I for one, Welcome our Robot Overlords

Chris O’Brien, Highly Valued

don’t forget to trust your gut, because you are the subject matter expert
example: multiplying two large numbers results in incorrect number
ChatGPT does not work that way
LLM is not a calculator; it’s meant to imitate human speech
why bother: better response times, faster resolution times, better customer satisfaction
output is fundamentally averages based on the data fed into it
given data and learn the associations
bad data turns into bad results
ChatGPT: is not meant to calculate, compute, we haven’t told it any rules
if it doesn’t know about something, it can’t tell you about it
there is no truly “average” customer, don’t have enough to build that profile, lack good coverage
results aren’t going to work for everyone
know the limitations
AI is just another tool in your toolbox

How to Build a Customer-Focused Engineering Escalation Process Everyone Likes

Ethan Walfish, LinkSquares

support responsibilities: serving customers, acting as voice of customer for other parts of business, handling transactional work during scale-up
types of bug dysfunction: all hands on desk (super expensive, doesn’t scale), growing backlog (black hole), only highest matters (only escalated gets resolved), hot potato (lack of clear ownership, bounce around)
how you may feel: support: help customers, fix bugs; engineering: ignore bugs, build features; overlap: prioritized work, lack of distraction, help the business
how do we get there? We triage and prioritize issues so engineering doesn’t have to. In exchange: we ask for clear responsibility zones, align on prioritization methods, responsivness when we indicate urgency
what do we need to do?
prioritization matrix: severity/priority 0-4
- severity 0 is a major incident, should be “pageable” issue, all hands on deck
- sev1: immediate work by eng-on-rotation
- sev2: work beings within next week by eng-on-rotation
- sev3: fix strategically
- sev4: fix strategically or may not get fixed
- Example: visibility (number of customers affected), business impact (unusable, hard block, inconvenient workaround, workaround, no business impact)
- Example: visibility; severity/failure type
responsibility map: probably have this, codify, make accessible, ideate on each team is responsible, get granular if needed, note EM and PM of each part
bug template: establish standards: give team something to follow, what information should be included, want actionable bugs, include logs/errors messages, expected vs actual results, consider naming structure, team receiving bugs has something to push back against
support medic role: support’s secret agent; want to ensure that Engineering is confusing on important issues and not user error, one of the bset tools is have a role that reviews new bugs and is responsible for triaging and handing off to engineering. Can be permanent, rotational. Should review/reproduce bugs o ensure not user-error or duplicates, sets severity
Handing off to Engineering: may want to ask for eng-on-call rotation, can be per team or org; limit the number that triage each bug; hand off should be different for different severities, product may want to get involved in this process
help desk automation: move to push vs pull, should not be incumbent on reps to check status of bugs manually, build integration points so when a bug is closed, help desk are reopened; views should empower your team to scale working on bugs, listen to their feedback; make sure to label/tag tickets with bugs
bug process: identify intake points, hand-off points
implement: matrix, responsibilities map, workflow diagram go right into new doc, one doc for support audience including medic, one doc for eng audience, recommend sharing new doc and running overview session with both teas and Q&A, then go implement and practice

The Intersection of Support and Community

Regina Walton

goal of support is extraordinary customer support
one way to do that is through community
hybrid approach: support and community, focus on driving support questions to community
community managers will respond, and community can help each other
community escalation process
answer what you can publicly
develop a game plan
document workflow
experiment and change as needed, use data

Overcoming Complexity as your Support Team Scales

Natasha Ratanshi-Stein, Surfboard

nailing support is a prerequisite for being able to scale a company
as company grows, so not complexity
complexity: multiple departments, owners, expand outside of geo, languages, routing, BPOs, seasonality, products, lines of communication, customers, more of everything
get the right tooling in place and invest in making them scalable to match complexity
set the right metrics that encourage the right behaviours: setting response times, monitoring performance at team/individual level, QA
maximize productivity without sacrificing quality: right automations and ticket volume/hand time analysis, feedback loop with product, internal tooling
hire the right people: have generalists and give them autonomy, can they get intimately familiar with your product? have they worked somewhere they have had to be deeply familiar with a catalogue of products or services? how are they at problem solving? how quickly can they think on their feet?
support ops to ensure support team is a profit centre: cost centre becomes a self-fulfilling prophecy, achieves retention and positive word of mouth
after more than one bad experience, 76% of consumers say they would rather do business with a competitor
principles for a profit support centre team:
1. Hire and train the right people. Generalists with autonomy.
2. Use the right tools. Invest in the right tools to orchestrate your team and enable them to execute.
3. Deploy thoughtful automation. Gather context faster, not to replace human interaction.
4. Track the right metrics. Measure outcomes not tickets.
5. Outsource only if you can handle it. Often means more work, not less. Training, QA, etc.
In the absence of this, it will become a bottomless pit of inefficiency, resentment, and cost.
ebook on Scaling a customer support team from Surfboard

End of day 1

Look forward to tomorrow!